yad.codes
yad.codes is my digital garden for the technical work: essays and research notes on coding agents, program synthesis and what the work costs, with the numbers pulled from the session logs. This page is its feed. The stories from the shows and the rooms are on the blog here; program synthesis research is on programsynthesis.pub.
-
What My $200 Codex Plan Was Actually Worth October 1, 2026
Seven weeks of Codex logs lined up against my weekly limit, the new Pro tiers, and the 62,500 credits that showed up the night after DevDay: what the pricing change actually means for how I use Codex.
-
Every Coding Agent Trend Is Program Synthesis September 10, 2026
Prompt engineering, context engineering, loop engineering, harness design, evals. If you know the source material, you can predict what gets named next.
-
Building StackUnderflow: Local-First Observability for Coding Agents May 19, 2026
A local-first toolkit that ingests session logs from 17 coding-agent providers to surface cost analytics, filesystem time-travel, and a session memory that agents can query mid-task. It answers two questions about your coding agents: what did this run actually cost, and what did I already learn here?
-
Coding Agents Are the Base Agent May 13, 2026
A practical mindset for picking up coding agents, even if you don't code. Destinations, maps, tools, and why every other agent is a coding agent in disguise.
-
Building Chimera: A Coding Agent Framework, Built by Coding Agents May 10, 2026
How building the same agent three times led me to decompose coding agents into composable primitives, and why I didn't write most of the code
-
58 Schmidhuber Papers in Pure NumPy with Claude Code May 9, 2026
58 Jürgen Schmidhuber paper stubs implemented in pure numpy via the same SPEC-driven agent-team workflow, with the build internals now grounded in raw session data
-
Sutro Yaro: Agent-Driven Research on Hinton's Problems May 3, 2026
How a SPEC issue, a wave of Claude Code agents, and GitHub PRs reproduced 53 Hinton experiments from 1981 to 2022 in one 49-hour run
-
Sutro Yaro: Agent-Driven Research on Energy-Efficient Learning March 14, 2026
How a study group in San Francisco turned coding agents into research agents by going back to 1960s AI problems with modern tools
-
Building Mycelium: A Digital Garden Theme for People Who Actually Write February 14, 2026
Why I built a digital garden theme around how I actually think and write
-
Building Bourbaki: A Math Agent That Can Check Its Own Work February 9, 2026
How I built an autonomous agent for mathematical reasoning with SymPy, Lean 4, and structured proof techniques, and what I learned along the way
-
Make an Honest Resume for Your Coding Agent February 8, 2026
Why giving your coding agent an honest map of what you know and don't know changes everything about how it helps you
-
Building Erdős Navigator: Claude Code as a Theorem Proving Assistant February 6, 2026
How program synthesis led me down a rabbit hole through HoTT and algebraic topology, and back out through LLMs and a toolkit for 1,179 unsolved math problems
-
The Missing Toolbox: Civil Engineering, Ethics, and Creative Writing Classes February 5, 2026
Why agent builders need lateral thinking from civil engineering, creative writing, and ethics
-
OWASP Top 10 for LLMs: A Pentester's Guide to Attacking and Defending January 29, 2026
How to not get pwned building with LLMs. A red teamer's take on the new attack surface.
-
Agent Sandboxes and the Proxmox Rabbit Hole January 24, 2026
How my obsession with home labs led me to understand why coding agents need isolation
-
Video Games Were Training Us for This January 22, 2026
How puzzles, strategy games, and chemical mixing in Resident Evil prepared me for coding agents
-
Coding Agent Cheat Codes January 20, 2026
CLI flags, environment variables, and proxy tricks for agentic engineering (or vibe coding, whatever you call it)
-
Coding Agents Made Me a Better Programmer January 16, 2026
Why AI coding tools forced me to think at a higher level of abstraction
-
Whether You're Vibing or Not, You're Going to Use Coding Agents January 16, 2026
Counterintuitive things after 125 active repositories and 50k+ commits in 2025
-
Red-Teaming GPT-OSS-20B: Lessons from the Kaggle Competition December 15, 2025
I tested 158 attack prompts against an open-source LLM and found a 27.2% success rate. Here's the methodology.